# Fast Classification of Large Time Series Datasets

arXiv: https://arxiv.org/abs/2312.06029 · HTML: https://mlsanity.com/paper/2312.06029/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 7 of its 21 references that are in the map.
- First posted on arXiv 10 Dec 2023.

## Builds on (most cited first, 7 of 7)

- [Deep learning for time series classification: a review](https://mlsanity.com/paper/1809.04356/) (2018)
- [InceptionTime: Finding AlexNet for Time Series Classification](https://mlsanity.com/paper/1909.04939/) (2019)
- [Time Series Classification from Scratch with Deep Neural Networks: A Strong Baseline](https://mlsanity.com/paper/1611.06455/) (2016)
- [LSTM Fully Convolutional Networks for Time Series Classification](https://mlsanity.com/paper/1709.05206/) (2017)
- [Adversarial Attacks on Deep Neural Networks for Time Series Classification](https://mlsanity.com/paper/1903.07054/) (2019)
- [Classification of Time-Series Images Using Deep Convolutional Neural Networks](https://mlsanity.com/paper/1710.00886/) (2017)
- [Bag of Recurrence Patterns Representation for Time-Series Classification](https://mlsanity.com/paper/1803.11111/) (2018)

## Abstract (arXiv)

Time series classification (TSC) is the most import task in time series mining as it has several applications in medicine, meteorology, finance cyber security, and many others. With the ever increasing size of time series datasets, several traditional TSC methods are no longer efficient enough to perform this task on such very large datasets. Yet, most recent papers on TSC focus mainly on accuracy by using methods that apply deep learning, for instance, which require extensive computational resources that cannot be applied efficiently to very large datasets. The method we introduce in this paper focuses on these very large time series datasets with the main objective being efficiency. We achieve this through a simplified representation of the time series. This in turn is enhanced by a distance measure that considers only some of the values of the represented time series. The result of this combination is a very efficient representation method for TSC. This has been tested experimentally against another time series method that is particularly popular for its efficiency. The experiments show that our method is not only 4 times faster, on average, but it is also superior in terms of classification accuracy, as it gives better results on 24 out of the 29 tested time series datasets. .